Flagellin-specific T cells induce colitis by recognizing antigen other than flagellin

This is very interesting study from Journal of Immunology (JI). Here, researcher led by Timothy Hand at the University of Pittsburgh Medical School, showed that CBir1 transgenic T cells, thought to cause gut inflammation by recognizing flagellin expressing microbiota, were in fact specific for non-flagellin antigens. 

CBir1 transgenic mouse have been used for past 10 years to model human Crohn's disease in mice. CBir1 tetramer+ T cells recognize epitope from flagellin, antigen thought to be a target in Crohn's disease. Interestingly, all of those experiments were done using CBir1 T cells on WT background. This could be an issue because it has been known for some time now that transgenic T cells on WT background could use alternative Vα chain to form TCR with new specificity.

To avoid such limitation, here the authors generated CBir1 mouse on RAG KO background that only expressed transgenic Vα:Vβ chains. Surprisingly, unlike CBir1:WT T cells, CBir1:RagKO T cells when transferred in immunodeficient hosts did not induce colitis, and it was not because CBir1:RagKO T cells differentiate into Foxp3+ T cells. It appears that CBir1:RagKO T cells just did not see flagellin epitope in adoptive host.

In vitro tests showed that both CBir1:WT T cells and CBir1:RagKO T cells could respond to DCs pulsed with flagellin epitope [it would have been better and more relevant here to use DCs pulsed with gut flora component].

Other tests showed that CBir1:WT T cells in gut tissue could express alternative Vα chain to form a completely new TCR specificity together with transgenic Vβ chain such as against Ags derived from Helicobacter (HH1713 tetramer). 

In summary, it appears that CBir1 T cells initiate colitis by recognizing non-flagellin antigen from the gut flora through non-CBir1 TCR and only following gut inflammation and gut leakage do flagellin-specific CBir1 transgenic T cells get activated and participate in overall colitis.

So, what is missing from this study? One, it would have been relevant to transfer in vitro activated CBir1:RagKO T cells or activate them in vivo directly to see if then they could initiate colitis. Second, the authors could have tried monocolonization of germ-free mice to see the source of non-flagellin microbiota. Third, there is inconsistency between Fig. 3B and Fig. 6A with regard of proliferation of  CBir1:WT T cells in response to Vanc-treated samples (in vitro it did not proliferate but in vivo it did).

posted by David Usharauli

Molecular mimicry to gut microbiota antigen protects against colitis but induces diabetes

Current issue of journal Cell has one very interesting but at the same time confusing research paper. In it, the authors proposed that

(a) diabetes susceptible mice strain, NOD, harbor CD8 T cells specific for microbiota antigen that cross-react with β cell antigen, IGRP, and

(b) such molecular mimicry prevents colitis but at the same time could induce diabetes.

First, the authors showed that MHC I alelle expressed in NOD mice (H2K^d) could bind IGRP_206-214 homologue derived from integrase family expressed by some gut Bacteroides species (BacIYL_36–44). 

At high dosage, such binding was functional in stimulating high affinity IGRP_206-214-specific T cells (17.4+ CD8 T cells).

Human T cells from PBMCs could apparently respond to it as well (though it is strange that it generated better stimulation index than Tetanus toxoid).

Then, the authors did the following experiment. They exposed IGRP^-/- 17.4+ TCR transgenic mice to chemical irritant (DSS) and observed that high affinity IGRP_206-214-specific T cells, 17.4+ CD8 T cells, but not low affinity ones (17.6+), could protect against colitis (I assume that they used IGRP^-/-mice to avoid diabetes development).

It appears that colitis protection depended on perforin expression by 17.4+ T cells. The authors speculated that 17.4+ CD8 T cells prevented colitis by eliminating dendritic cells laden with microbiota-derived antigen (BacIYL_36–44).

As a confirmation, the authors showed that germ-free TCR Tg NOD mice colonized with Bacteroides species expressing BacIYL_36–44 were protected against colitis.

Colitis protection was observed even in classical, adoptive naive CD4+ T cell transfer colitis model.

Interestingly, however, transfer of T cells from pre-diabetic NOD mice into germ-free NOD.scid mice colonized with Bacteroides species expressing BacIYL_36–44 did not accelerate diabetes development (here I assume DSS is required to accelerate T cells priming against IGRP by creating dysbiosis).

In summary, this study suggests the following scenario: diabetes-inducing CD8+ T cells cross-react with gut microbiota-derived antigen. When such microbiota-derived antigens become visible to T cells (during dysbiosis?) CD8+ T cells migrate to gut and eliminate dendritic cells laden with cross-reactive antigens. By eliminating DCs, other T cells are not able to induce inflammation in the gut, thus no colitis. However, the same beneficial CD8+ T cells later migrate to β cells, recognize similar looking antigen, IGRP, and mediate its destruction and diabetes.

Does such circuit makes any evolutionary sense? 

Update: Interestingly, other research group previously detected different set of gut microbiota antigens cross-reactive to IGRP_206-214. They used TCR NY8.3 transgenic NOD mice (that recognize the same IGRP epitope) and found that these CD8 T cells cross-reacted with IGRP_206–214 homologous peptide, W15944, derived from L. goodfellowii, a member of the phylum Fusobacteria (gram-negative anaerobe), a human and NOD mouse oral commensal. 

posted by David Usharauli   

Protozoa-enabled non-genetic colitis after T cell transfer in immune-deficient mice

Adoptive transfer of naive T cells into T-cell deficient host mice has been used as a colitis (gut inflammation) model that led to discovery of FOXP3+ T regulatory cells which when co-transferred with naive T cells prevented gut immunopathology.

A new study in Journal of Experimental Medicine, however, provided evidence to show that there is a limit how much FOXP3+ T regs could do. This study found that adoptive transfer of naive T cells into Rip2−/−Rag1−/− mice (RIP2 is an essential signaling adapter molecule downstream of both NOD1 and NOD2) led to protozoa, Tritrichomonas muris-enabled dominant colitis that could not be prevented by FOXP3+ T regs.

Initially, the authors observed in non-littermates that naive T cell transfer led to severe gut inflammation in  Rip2−/−Rag1−/− host as compared to just Rag1−/− mice [though it is strange that in their mouse facility Rag1−/− host did not show weight loss after T cell transfer].

Interestingly, colitis in Rip2−/−Rag1−/− hosts could not be prevented by co-transfer of FOXP3+ T regs.

Co-housing and littermate control control experiments confirmed that a non-genetic factor was responsible for colitis development in Rip2−/−Rag1−/− hosts and that factor could be transferred between mice when co-housed together.

Fecal matter analysis showed that Rip2−/−Rag1−/− hosts were selectively enriched with protozoa Tritrichomonas muris, and that its transfer to other mouse accelerated T cell-mediated colitis.

In summary, this study shows that Rip2−/−Rag1−/− double deficient hosts harbor protozoa Tritrichomonas muris that by itself or through modulation of gut microbiota establishes a dominant colitogenic gut ecosystem that is even transferable to genetically non-related mice.  

David Usharauli

  

Division of labor among thymus-derived Foxp3+ regulatory T cells

This week Nature Immunology published study that showed another level of "division of labor" among regulatory Foxp3+ T cells (Tregs). The authors revealed presence of two types of thymic Tregs defined by expression of three receptors, GITR, PD-1 and CD25 (GITR^hiPD-1^hiCD25^hi and GITR^loPD-1^loCD25^lo Tregs cells) that displayed non-overlapping functionality.

Specifically, only GITR^loPD-1^loCD25^lo Tregs cells prevented colitis development in adoptive transfer experiment by converting responding naive T cells into induced Tregs.

On the other hand, only GITR^hiPD-1^hiCD25^hi Tregs cells prevented uncontrolled proliferation of endogenous T cells when transferred into Treg-depleted host.

In summary, this study revealed that even among thymus-derived natural Tregs there is a division of labor. It is possible that difference between these two types of thymic Tregs is also related to their differential migration pattern as it was suggested by one recent study on KLF2.

David Usharauli

CARD9 modulates gut inflammation via microbiota-specific tryptophan metabolism

In my previous blogpost I discussed new Science paper that showed relationship between autophagy pathway (ATG16L1) and microbiota-secreted product (PSA-OMV) in driving Crohn's disease (CD) risk. 

As it happens frequently in scientific publication, yesterday another prestigious journal Nature Medicine published another study from another research group that showed relationship between Card9 (C-type lectin sensor, also involved in NOD2 signaling) and microbiota-specific tryptophan metabolism in driving inflammatory bowel disease (Crohn’s disease and ulcerative colitis) risk.  

It is known that Card9^−/− mice are more susceptible to colitis. Here, the authors noticed that Card9^−/− mice are slow to recover after dextran sulfate sodium (DSS)-induced colitis (a self-limiting colitis model) and expressed fewer IL-22⁺ cells, a cytokine with well-known beneficial effects on intestinal homeostasis.

Examination of composition of the fecal bacterial microbiota using 16S rDNA sequencing revealed differences between WT and  Card9^−/− mice gut flora.

Interestingly, germ-free mice transplanted with gut flora from Card9^−/− mice were more susceptible to DSS-induced colitis. This indicated that unlike WT microbiota, Card9^−/− mice microbiota failed to provide "healing" signaling to host's gut epithelium.

One mechanism for this "healing" could be the modulation of aryl hydrocarbon receptor (AhR) activation. Tryptophan can be metabolized either by the gut bacteria into indole derivatives (i.e IAA) or by host cells into kynurenine (Kyn) via indoleamine 2,3-dioxygenase 1 (IDO1). Indole derivatives are AHR ligands and promote IL-22. Indeed, WT microbiota, but not Card9^−/− microbiota, promoted AhR ligand production (Of note, exogenous IL-22 could normalize AhR ligand production and colitis susceptibility in Card9−/− mice).

In vitro assay with AhR reporter system confirmed that Card9^−/− microbiota were defective in activation of AhR signaling.  

Furthermore, supplementation with Lactobacilli strains that are able to produce AhR ligands could restore "healing" effect of Card9^−/− microbiota on gut epithelium.

Finally, the authors showed that fecal samples from IBD patients were indeed deficient in promoting AhR signaling via indole products (genotyping confirmed that fecal samples from IBD patients with Card9 risk allele was associated with reduced AHR activation in their in vitro reporter assay). Of note, the authors said that no such association was observed among other major IBD risk alleles, such as NOD2, ATG16L1 and LRRK2.

In summary, this new study suggests that alterations in IBD risk genes modifies microbiota composition that in turn could tip the balance in favor of "non-healing" microbiota resulting in reduced level of "healing"AhR ligands.

David Usharauli